Table 3.

Performance of MTD selection of designs across scenarios when maximum sample size is 60.

	Simulation scenario
	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
Design	Selection of correct MTD $(S_C)$
I2D	0.43	0.71	0.41	0.9	0.86	0.34	0.59	0.11	0.15	0.43	0.47	0.47	0.23	0.25	0.24
I2D.pq	0.58	0.73	0.54	0.81	0.88	0.35	0.54	0.2	0.09	0.3	0.58	0.45	0.28	0.4	0.16
Copula	0.53	0.6	0.52	0.1	0.9	0.44	0.65	0.15	0.32	0.26	0.56	0.45	0.12	0.12	0.42
Copula.pq	0.59	0.6	0.58	0.06	0.91	0.37	0.55	0.4	0.12	0.01	0.53	0.26	0.4	0.05	0.13
Hierarchy	0.61	0.64	0.62	0.61	0.81	0.45	0.45	0.3	0.47	0.52	0.6	0.28	0.47	0.24	0.52
Hierarchy.pi	0.61	0.5	0.71	0.32	0.78	0.42	0.52	0.42	0.09	0.11	0.6	0.24	0.39	0.11	0.17
POCRM	0.75	0.71	0.69	0.78	0.54	0.59	0.56	0.59	0.52	0.58	0.74	0.52	0.46	0.57	0.48
POCRM.skeleton	0.74	0.67	0.71	0.8	0.73	0.57	0.51	0.6	0.49	0.53	0.77	0.5	0.44	0.54	0.35
DFCOMB	0.54	0.76	0.66	0.65	0.54	0.33	0.69	0.48	0.15	0.47	0.44	0.63	0.37	0.18	0.67
DFCOMB.pq	0.69	0.8	0.65	0.64	0.52	0.34	0.71	0.3	0.09	0.14	0.57	0.56	0.33	0.35	0.45
DFCOMB.cut	0.59	0.79	0.64	0.65	0.54	0.26	0.70	0.36	0.24	0.57	0.51	0.63	0.36	0.15	0.60
DFCOMB.sensitive	0.54	0.78	0.66	0.62	0.58	0.33	0.69	0.44	0.15	0.47	0.46	0.61	0.38	0.20	0.64
gCRM	0.69	0.65	0.71	0.42	0.81	0.59	0.67	0.34	0.47	0.55	0.64	0.47	0.49	0.17	0.48
gCRM.pi	0.58	0.64	0.61	0.47	0.81	0.5	0.74	0.26	0.48	0.51	0.61	0.57	0.34	0.22	0.23
cBOIN	0.7	0.69	0.7	0.62	0.72	0.58	0.74	0.38	0.4	0.45	0.75	0.57	0.38	0.4	0.37
cKeyboard	0.67	0.7	0.7	0.6	0.72	0.56	0.71	0.38	0.4	0.45	0.73	0.58	0.38	0.43	0.36
bCRM	0.72	0.75	0.66	0.76	0.52	0.51	0.63	0.51	0.37	0.5	0.62	0.54	0.39	0.35	0.47
bCRM.skeleton	0.75	0.72	0.73	0.84	0.69	0.59	0.64	0.51	0.36	0.36	0.69	0.52	0.47	0.33	0.45
bCRM.cut	0.75	0.67	0.71	0.71	0.59	0.56	0.63	0.54	0.40	0.52	0.71	0.53	0.40	0.37	0.43

Notes: I2D: design I2D with parameter $p_j$ and $q_k$ specified in ‘Main setting’ of Table 1.

I2D.pq: design I2D with parameter $p_j$ and $q_k$ specified in ‘Alternative setting’.

Copula: design Copula with parameter $p_j$ and $q_k$ specified in ‘Main setting’.

Copula.pq: design Copula with parameter $p_j$ and $q_k$ specified in ‘Alternative setting’.

Hierarchy: design Hierarchy with ${\pi }_{1k}$ and ${\pi }_{j1}$ specified in ‘Main setting’.

Hierarchy.pi: design Hierarchy with ${\pi }_{1k}$ and ${\pi }_{j1}$ specified in ‘Alternative setting’.

POCRM: design POCRM with skeleton specified in ‘Main setting’.

POCRM.skeleton: design POCRM with skeleton specified in ‘Alternative setting’.

DFCOMB: design DFCOMB with $p_j$ and $q_k$, and escalation/de-escalation probability cutoff specified in ‘Main setting’.

DFCOMB.pq: design DFCOMB with $p_j$ and $q_k$ specified in ‘Alternative setting’, and escalation/de-escalation probability cutoff specified in ‘Main setting’.

DFCOMB.cut: design DFCOMB with $p_j$ and $q_k$ specified in ‘Main setting’, and escalation/de-escalation probability cutoff specified in ‘Alternative setting’.

DFCOMB.sensitive: design DFCOMB with $p_j$ and $q_k$, and escalation/de-escalation probability cutoff specified in ‘Main setting’, but target toxicity interval boundaries suggested by one of our reviewers.

gCRM: design gCRM with ${\pi }_{1k}$ and ${\pi }_{j1}$ specified in ‘Main setting’.

gCRM.pi: design gCRM with ${\pi }_{1k}$ and ${\pi }_{j1}$ specified in ‘Alternative setting’.

bCRM: design bCRM with skeleton, and escalation/de-escalation probability cutoff specified in ‘Main setting’.

bCRM.skeleton: design bCRM with skeleton specified in ‘Alternative setting’, and escalation/de-escalation probability cutoff specified in ‘Main setting’.

bCRM.cut: design bCRM with skeleton specified in ‘Main setting’, and escalation/de-escalation probability cutoff specified in ‘Alternative setting’.